[Xen-bugs] [Bug 1480] New: Solaris HVM domains with multiple VCPUs see fatal pagefaults that aren't.

[bugzilla-daemon@xxxxxxxxxxxxxxxxxxx]

http://bugzilla.xensource.com/bugzilla/show_bug.cgi?id=1480

           Summary: Solaris HVM domains with multiple VCPUs see fatal
                    pagefaults that aren't.
           Product: Xen
           Version: unstable
          Platform: x86-64
        OS/Version: other
            Status: NEW
          Severity: normal
          Priority: P2
         Component: HVM
        AssignedTo: xen-bugs@xxxxxxxxxxxxxxxxxxx
        ReportedBy: frank.vanderlinden@xxxxxxx


In our 3.3.2-pre - based tree, we are seeing problems with Solaris HVM domains
getting fatal pagefaults, while the guest page tables look just fine.

This issue is reproduced by running (installing or running a stresstest) a
Solaris HVM domain with multiple VCPUs. We've reproduced the issue with > 2
VCPUs.

What happens is that the Solaris kernel allocates some kernel memory, and then,
shortly afterwards, touches it (e.g. to zero it out or otherwise). This
generates a fatal pagefault (page not present). However, inspecting the guest
page tables shows that they are fine. The page is mapped as far as the guest is
concerned.

This points to an issue with the shadow page table code, probably relating to
domains with multiple VCPUs.

An additional data point is that enabling hardware assisted paging makes the
problem go away on a dom0 system that otherwise reproduces the problem in
guests reliably.


-- 
Configure bugmail: 
http://bugzilla.xensource.com/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

_______________________________________________
Xen-bugs mailing list
Xen-bugs@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-bugs